Professional Documents
Culture Documents
Administração
Contemporânea
Journal of Contemporary Administration e-ISSN: 1982-7849
Methodological Article
ABSTRACT RESUMO
Objective: the literature on board interlocks has increased in recent years, Objetivo: pesquisas sobre board interlocks vêm crescendo nos últimos anos,
focusing on understanding board composition and its relationships with com foco no entendimento da composição dos conselhos assim como suas
other companies’ boards. Such studies usually require multiple procedures relações com conselhos de outras companhias. Esses estudos normalmente
of data extraction, handling, and analysis to create and analyze social requerem múltiplos procedimentos de extração, tratamento e análise de dados
para a criação e análise das redes sociais. Entretanto, esses procedimentos não
networks. However, these procedures are not standardized, and there is
são padronizados, havendo uma falta de estudos metodológicos com instruções
a lack of methodological instructions available to make this process easier para tornar este processo mais simples. Assim, este tutorial pretende descrever
for researchers. This tutorial intends to describe the logical steps taken to a sequência lógica de passos a serem percorridos para realização da coleta de
collect data, treat them, and map and measure the network properties to dados, tratamento, mapeamento e análise das redes sociais, para prover aos
provide researchers with the sources to replicate it in their own research. pesquisadores os insumos necessários para replicação desses procedimentos
We contribute to the literature in the management field by proposing an em suas pesquisas. Nesse sentido, este tutorial contribui com a literatura
empirical methodological approach to conduct board interlocks’ research. no campo de pesquisa da administração por propor uma metodologia para
Proposal: our tutorial describes and provides examples of data collection, condução de pesquisas em board interlocks. Proposta: o tutorial descreve e
directors’ data treatment, and the use of these data to map and measure exemplifica a extração e tratamento dos dados das empresas e seus conselheiros,
o uso destes dados no mapeamento das redes de board interlock e a medição
network structural properties using an open-source tool — R statistical
de suas propriedades estruturais, utilizando uma ferramenta open source, o
software. Conclusions: our main contribution is a tutorial detailing the software estatístico R. Conclusões: nossa principal contribuição é fornecer
steps required to map and analyze board interlocks, making this process um tutorial que orienta o processo de mapeamento e análise dos board
easier, standardized, and more accessible for all researchers who wish to interlocks, tornando-o mais acessível aos pesquisadores que desejam adotar esta
develop social network analysis studies. abordagem de pesquisa.
Keywords: board interlocks; social network analysis; tutorial. Palavras-chave: board interlocks; análise de redes sociais; tutorial.
1st round
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 1
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 2
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
networks of board members appointed by state-owned board interlocks, SNA can be used as a framework to
enterprises and/or institutional investors. understand relationships and behaviors across many other
research subjects. For instance, in management it can be
A social network is a collection of individuals or
used to evaluate the relationship between countries through
companies interconnected by many different sorts of
trading patterns (Smith & Sarabi, 2021a), and in medical
links (Lazzarini, 2008). SNA can be used to analyze the
studies, by evaluating the social interactions across animal
consequences or causes of board interlocks existence. One
species (Rocha, Ryckebusch, Schoors, & Smith, 2021).
of the challenges faced by researchers is to obtain access to a
To illustrate the process focused on this paper, we used a
detailed dataset related to board members and companies.
database of Brazilian listed companies in 2019 and 2020 to
Usually, these datasets are not easily manageable. In some set out the networks and measure their properties, using R
countries, such as Brazil, the data for public companies Studio to apply the steps described. The sample resulted in
is non-structured, so researchers must follow many steps 592 companies listed in 2019 and 629 companies in 2020.
in the performance of their research, which require a lot
of hand-collected data. Another drawback in this field This article is organized as follows: we first present
of research is that the procedures to create and analyze the literature concerning board interlock and SNA; the
networks’ measures are usually not well detailed in the following section describes how to measure interlocks
previous studies. This can be a result of word limitations in our dataset, following the steps of data collection and
or standards from different fields of study and journals. treatment, forming a relational matrix, and then visualizing
Therefore, it is hard to find some standard or guidance and measuring board interlocks using R; finally, we explore
when conducting social network analysis research. In this some of the advantages and limitations of the method and
tutorial, we aim to explain with details, step by step, the present our concluding remarks.
procedures involved in the conducting of data collection,
data treatment, and how to map and measure the network BOARD INTERLOCKS LITERATURE
properties of board interlocks. This study contributes
to the literature in several ways: (a) by establishing a Board interlock is the relationship between two
methodological way to conduct procedures and handle company boards when they share at least one director.
social network analysis; (b) by providing researchers In other words, board interlock occurs when a director
with practical instructions to properly conduct these of company ‘A’ also belongs to the board of directors of
procedures, spending less time in manual tasks, increasing company ‘B’ (Fich & White, 2005). Besides this, a board
data reliability; and (c) providing a standard to conduct interlocking network is a group of company boards and
such tasks. This study can also encourage more masters all the interlocks among them (Wong et al., 2015). Board
and PhD students to explore the SNA and board interlocks interlock can include studies in different fields, such as
fields of study, not yet widely explored in Brazil. finance, governance, and sociology.
Furthermore, this tutorial can be of interest of Interlocks can be seen as a communication tool
researchers analyzing the “G” of the recent ESG topics, between different companies. Lamb and Roundy (2016)
including the causes and consequences of the relationships showed that interlock research could adopt two perspectives:
between different actors (e.g., companies and directors). the company’s perspective and the director’s perspective.
Within the corporate governance context, the method Interlocks can occur from a company standpoint to create
proposed in this tutorial makes it possible to analyze the links between companies, reach financial objectives,
composition of the networks, encompassing relevant issues reduce environmental uncertainty, or improve the
such as diversity and connectivity of boards. For example, board monitoring ability. In addition, companies can be
we may investigate if while pushing the gender diversity interlocked as a signal to current and potential investors.
agenda companies become more connected as the presence For example, when a company appoints a director from
of women increases, or alternatively may be resulting in the a corporation with a high reputation, this can signal its
‘super-connected director’ phenomena (Crabtree, 2011). quality. From the director’s perspective, they found that
In addition, our tutorial may allow researchers to evaluate the directors want to be interlocked to advance their careers
how connections between companies from different sectors and extend their connections by making social links.
affect different aspects of corporate governance, including
Board interlock studies have been increasing over
ESG requirements and policies (Harjoto & Wang, 2020).
time. Many try to find evidence of board interlocking
Additionally, we briefly discuss SNA concepts consequences, such as the effects of board interlocking
and the ideas presented in the most high-impact studies on strategic decision-making, such as the design of
available in this field, shedding light to possibilities of CEO compensation package (Fich & White, 2003;
studies using SNA. Even though in this study we explore Larcker, Richardson, Seary, & Tuna, 2005), earnings
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 3
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
management (Tham, Sultana, Singh, & Taplin, relationships for a given network, which, in SNA language,
2019), company performance (Vesco & Beuren, is a network graph or sociogram (Scott, 1992).
2016), and the strength of corporate governance
(Silva, Silva, Vasconcelos, & Crisóstomo, 2019). Bebchuk Cohesion and density
and Fried (2004) also showed that the monitoring
function could fail when a director serves several The most relevant structural properties in the SNA
boards. They would be more likely to agree with CEOs’ for a network are characterized as cohesion, involving
opportunistic behavior or other situations in which such elements as density, centralization, and centrality
the shareholder’s profit maximization could be injured. (Borgatti, Everett, & Johnson, 2018; Scott, 1992). Density
On the other hand, Mazzola, Perrone, and Kamuriwo is a property at the network level indicating how compact
(2016) argued that interlocked companies are more likely or cohesive the network is. It is obtained by measuring
to achieve new product development outcomes due to the general level of connections between all the participant
providing experience sharing and information to mitigate nodes and is used for comparing different networks (Kilduff
drawbacks. By this view, board interlock could accelerate
& Tsai, 2003). In quantitative terms, it corresponds to the
the development process of new products.
number of connections and is expressed as a proportion
Each country has its laws designed to regulate board over the number of possible links (Borgatti et al., 2018).
interlocks. In Brazil, Federal Law 6,404/76 has established It can vary from zero to one, reaching the value ‘1’ when
that a director who sits on a company’s board may not be there are connections between all the participating nodes
elected to sit on a concurrent company board (Lei n. 6.404, (Wasserman & Faust, 1994). For example, Croci and Grassi
1976). However, such an arrangement is not prohibited. (2014) measured the network density of the companies
The ruling is often different in other countries such as the listed on the Italian Stock Exchange, obtaining the value of
U.S., where the Clayton Act (1914) restricts directors from 0.0268 for the main component. In that paper, this value
sitting on two boards of competitor companies. Besides is associated to a tendency of companies to be strategically
this, public companies in Brazil do not have to directly connected.
disclose the interlocks between their directors. Hence, in
such a case, to identify interlocks between companies, it However, density does not state how the links are
is necessary to obtain the board composition from the distributed across the network. Other cohesion measures
Reference Form, which consists of a set of information that that help in this respect are the number of components, their
public companies need to deliver to CVM annually. There size, and the degree of connectedness or fragmentation.
is a range of information available in this document, such The larger the main component (component with the
as board composition, board and CEO features, as well as higher number of nodes in the network), the greater the
several financial and economic data. As we will explain in number of nodes that are part of this component and the
the following sections, we then need to perform procedures greater the global cohesion of the network (Borgatti et al.,
on the database extracted to identify interlocks. This kind 2018). Connectedness is the proportion between pairs of
of database makes it possible to map the network formed nodes that can reach each other by a certain path, while
between public companies and directors and analyze the fragmentation measures precisely the opposite, that is, the
existent patterns of relationships. An effective way of doing ratio between pairs of unreachable nodes (Borgatti et al.,
this is using social network analysis tools. 2018). The connectedness is calculated using Equation
1, and the fragmentation is one minus connectedness
SOCIAL NETWORK ANALYSIS (SNA) (Borgatti et al., 2018).
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 4
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
Centrality measures where N is the number of nodes and d(i,j) is the function
of the distance between i and j. In a study evaluating the
The centrality property allows identification of the performance of Chinese foundations, for example, a high
importance or ‘degree of popularity’ of a particular node level of closeness centrality of the board interlocking of
(Scott, 1992), therefore, a property at the node’s level. From these institutions appeared as a factor that affects positively
the perspective of the position of the nodes in the networks, the outcomes of income and public welfare expenditure
we can observe three essential types of centralities: degree (Wu, Zhang, & Chen, 2021).
centrality, closeness centrality, and betweenness centrality.
These metrics reflect the importance of individual nodes Finally, betweenness centrality highlights nodes that
in a network under different aspects (Wasserman & Faust, connect different nodes or groups of individuals in the
1994). network (Wasserman & Faust, 1994). It can explain the
power of influence a node has in a network because it joins
Degree centrality is that central node with the other nodes or groups isolated amongst the other nodes in
most links with other nodes in the network (Lazzarini, a network (Zhu, Watts, & Chen, 2010). We calculate the
2008). According to Zhu, Watts, and Chen (2010), degree betweenness centrality of node i using Equation 4.
centrality measures the number of links with other nodes
in the network. In business power studies, for example, the
literature points out that those nodes with a high degree (4)
of centrality generally exercise dominance over more
peripheral nodes (Kilduff & Tsai, 2003). By the adjacency
matrix X of a network, degree centrality is a simple sum
of rows or columns of the adjacency matrix. If di is the where gjik is the number of ways connecting j and k through
degree centrality of node i and xij is the (i, j) record of the i, and gjk is the total number of ways connecting j and k.
adjacency matrix, it must be calculated as in Equation 2 Using again the example of the Italian Stock Exchange
(Borgatti, Mehra, Brass, & Labianca, 2009). companies (Croci & Grassi, 2014), the minimum and
maximum values measured for betweenness centrality were
respectively 0 and 0.1506. Pirelli & C. was the company
(2) with the higher score, thus being in a position to receive
and distribute a high volume of information.
The study of interlocked boards using SNA involves
In the example of the Italian companies (Croci & a particular network analysis mode, called two-mode data
Grassi, 2014), the average degree centrality measured for or affiliation data (Borgatti et al., 2018). There are two types
the main component was 5.4879, ranging from 1 to 32, of actors: the directors (the first mode), and the events,
being Pirelli & C. the company with the highest degree which are the boards of affiliated companies (second mode).
centrality and so the most influent and powerful one. The literature usually defines the companies/boards as the
Closeness centrality measures how close a node is nodes of the network and the links between the companies
to all the other nodes in the set of nodes (Wasserman & as the interlocks (Scott, 1992).
Faust, 1994). We can interpret closeness centrality as the
minimum time the flow of information needs to arrive
INVESTIGATING BOARD INTERLOCKS:
at another node in the network (Borgatti et al., 2009).
Freeman, Roeder, and Mulholland (1979) posit that A TUTORIAL ON BRAZILIAN LISTED
closeness centrality is the distance from a node to all the COMPANIES
others, where the smaller values identify highly central
nodes. Researchers often use a normalized version, dividing We take certain essential steps to measure the
each node centrality into n-1, reversing the values so that interlocks between directors and companies in the Brazilian
large numbers correspond to greater centrality, ranging market: first, we perform the data collection procedures
from zero to one. Thus, the normalized closeness centrality from a specific package on R software, with some searching
of node i will be represented by Equation 3. parameters. Second, the data are treated using automatic
methods on R. Third, we build the adjacency matrix to
identify companies’ relationships through the sharing of
(3) at least one director. Finally, we map the network and
measure it using R. A summary of the steps is provided in
Figure 1.
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 5
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
As an empirical example of our method, we used a the board of directors’ composition and its connections,
publicly listed dataset of Brazilian companies covering 2019 we will select only the data we need in #Step3. To get
and 2020. Each step required to conduct the procedures is only the members of the board of directors, we filter the
described over the course of this paper, and these procedures type of director, setting the code.type.job variable of the
can be replicated using the code available. df_board_composition.
The result of #Step3 is a table with the company
Collecting and treating Brazilian name, CNPJ (Cadastro Nacional de Pessoa Jurídica), the
companies’ data (#Step1 to #Step5) corporate registration number with the Brazilian tax
authority (Receita Federal do Brasil), company name,
The first step involves the collection of corporate reference date (information date), CVM code, director’s
data from the Reference Form (FRE). These data are name, and CPF (individual taxpayer registration) number,
available on the CVM website. Although these are publicly which is a document provided by the Brazilian tax authority
available data, it would have been necessary to extract them to identify Brazilian citizens and foreign citizens resident
individually for each company if we had to use the website. as taxpayers in Brazil. As the companies fill out the form
To collect the corporate data on a range of companies, we on FRE manually, there is a strong likelihood of mistakes
used the GetFREData package (Perlin et al., 2018) from existing in the information. Usually, we can find some
R statistic software. The additional packages used in this errors in FRE, such as CPFs with less than 11 digits or even
tutorial are described in our code, and these packages can CPF information missing entirely. The lack of standardized
be replicated with script “1” (#packages and libraries). information could interfere with our analysis. Therefore, we
#Step1 of our code allows the researcher to collect data on provide data treatment procedures to mitigate this risk and
Brazilian public companies available on the FRE, by simply correct the original database’s common mistakes.
selecting the period requiring analysis. However, connection First, we standardize the names and CPFs of the
problems or high latency can interrupt the download. In directors using #Step4. After this, we take those cases where
these cases, the program will present a warning such as “Try the CPF information is missing or incorrect and search for
rerunning the code as the corrupted zip file was deleted each director’s correct CPF in other entered cases. Finally, if
and will be downloaded again.” The user then needs to run we could not find the CPF, we create a code for this director
#Step1 again, and this time the program will check the instead of the CPF to keep a unique code for each director
companies already downloaded and jump to a download of in our database. To do this, we run #Step5. These procedures
the remaining companies. The files relating to companies are essential to be able to correct errors in the dataset and to
that have already been downloaded will not be downloaded guarantee its reliability.
again because the files have been saved in the cache.
Moreover, data collection could take a while because Creating the adjacency matrix using R
it downloads and reads each company’s FRE files. We Studio (#Step6 and #Step7)
therefore provided a ready dataset like the one created by
running #Step1 in .rds format (R native database). This To illustrate some possibilities for the application of
dataset is available to import .rds file in #Step2, choosing the SNA to board interlocks, we used IGraph package (Csardi &
I_fre.rds file that is available in the attachments to this paper. Nepusz, 2006) to generate the graphs in R, which provides
This was done to optimize the length of time taken for data graph characteristics and manipulates a large and complex
collection. We collected this database while this paper was network (Meghanathan, 2017). We used the SNA package
being created, meaning some data could be outdated. To (Butts, 2008) to generate and analyze the network measures
reach the updated database, we suggest running #Step1. and Keyplayer package (An & Liu, 2016) to generate the
Once the data collection process has finished (on fragmentation measure.
#Step1 or #Step2), we have a list containing all of the The relational matrix allows us to identify the
companies’ available corporate FRE data. As we focus on relationship between two or more companies that share at
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 6
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
least one director. Each matrix cell shows whether a node The outcome is shown in Figure 2. This visual comparison
A is interconnected with a node B (Lazzarini, 2008). For did not show significant change in the general configuration
instance, we create two relational matrices, one for each year of our sample’s network. The pattern of relationship among
of our sample, 2019 and 2020. First, we create a matrix with the companies has remained relatively similar from one
the company and directors, using companies and directors’ year to the next. Both sociograms represent relatively low
previously created codes. This matrix identifies each director density, high fragmented networks. Part of the companies
and enters the information ‘1’ when the director sits on the are connected forming a great group, other parts constitute
respective company’s board of directors. We then multiply
small subgroups of two (diads), three (triads), or more
this matrix by its transposed matrix to establish a ‘companies
companies and the rest of the companies are isolated, not
x companies’ square matrix. In this new matrix, we can
identify the number of directors that companies share at that connected to any other company.
time. The diagonal matrix will be entered as zero because it Now, we finally have the relational matrix and the
means the relationship between the company and itself. This networks ready. We can also export the matrix in CSV format
process is detailed in #Step6 for each year that we want to by using #Step7, which can be used to map and generate
create the relational matrix. the network measures in software different to R. The matrix
We can also plot the network graphs for each year of created is shown in Figure 3. We will then present the steps
our sample in the same step, using the commands described. for measuring and visualizing networks using R.
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 7
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
Visualizing and measuring board interlocks each sample company. A sample of these results is shown
using R Studio (#Step8 to #Step10) in Table 2, showing where we selected specific companies
to illustrate how the data is shown. These results can be
Our empirical example generates measures at the exported for analysis and included in the econometric
network’s level: size, density, degree of centralization, models as interest, independent, or moderating variables,
betweenness centralization, closeness centralization, according to the needs of each study.
connectedness, and fragmentation. In #Step8 we create These network measures could be used as variables in
a data frame with these measures for each year, making econometric models. To do this, we join the measures for
comparisons and analyses. The results are shown in Table all years of the sample. In Table 3, we show the descriptive
1. In the same step, we create a data frame with only the statistics of the consolidated sample for each variable. This
centrality measures (degree, betweenness, and closeness) for procedure is described in #Step9.
Taking a closer look at ‘degree’ in Table 3, the The minimum number of directors companies share is one
average of interlocks established between companies is (because we are considering only the interlocked companies)
6.52. Meanwhile, the median is four, which means that and the maximum is 32. Coincidently, this is the same
the interlocked companies usually share four directors. maximum number as in Croci and Grassi (2014). After this,
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 8
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
we can generate the graphs based on the measures in #Step9. other companies, directly or indirectly (Figure 3). Like in
The first is degree centrality, which shows the nodes that are Figure 2, the visualization of these sociograms permits to
more central in the network. Our empirical example shows conclude that the general pattern of relationship of the
the more prominent companies in the Brazilian market network has remained from 2019 to 2020, with a significant
environment in a bigger size, considering the amount of number of central companies. However, some changes
board interlocking between them. could be happening on companies that are more central.
In this case, the more prominent companies are This identification is facilitated using the degree centrality
those that have the greatest number of connections with measure (see Table 4).
Following, running #Step10, we can also generate companies have a lower degree of centrality in the network by
the histogram of degree centrality. It shows the distribution sharing from zero to five directors (the more peripheral ones).
of degree centrality in the companies of our sample. We can Few companies share more than 20 directors in our sample in
see from our empirical example that in both years, most both years. The histogram is illustrated in Figure 5.
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 9
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
Again, the graphs for 2019 and 2020 are similar, with represent changes in degree centrality to top companies.
few differences in the distribution of the number of directors We can see that the top seven central companies remain the
shared by the companies. Variations that are more significant same, but important changes happened with the emergence
probably could happen comparing longer periods. of company 18139 and loss of prominence of companies
21822, 21857, and 00906. Company 18139 is now in
Another possible analysis regarding the degree position to share information and experiences with a large
centrality involves a comparison of the companies with the number of other companies, and then obtain benefits such
highest degree of centrality to identify eventual changes. To as better strategic decisions (Fich & White, 2003; Larcker et
exemplify this, we listed companies with a degree centrality al., 2005) and the acceleration of the development of new
higher than 24 (Table 4). Companies marked in bold products (Mazzola, Perrone, & Kamuriwo, 2016).
Our third graph (Figure 6) presents the betweenness this privileged position and that several other companies
centrality, showing in a bigger size the companies that lost this privilege due to changes in the composition of
act as a bridge in the network by connecting different the companies’ boards. This privileged position can imply
companies. A simple comparison of the sociograms makes that companies with higher betweenness centrality could
it possible to verify that there was an increase in the number control the flow of information in the network. As long
of companies with high intermediation centrality. That is, as this structure changes over time, information flow can
there are more companies in a privileged position, being also change.
able to influence or restrict the interactions between other
Finally, we created the normalized closeness graph
companies and obtain benefit from that (Zhu et al., 2010).
(Figure 7), showing the lower distances from one node
As in the case of degree centrality, it is interesting to all the others. A high level of closeness indicates that
to compare the companies with the highest values of a node is highly central. Our sample shows that many
betweenness centrality to identify eventual changes. In companies are central in the network since there are many
Table 6, we listed the first 15 companies with the highest big-sized nodes. It means, for example, that companies
betweenness centrality each year. Companies marked with high values of closeness centrality potentially receive
in bold represent the changes observed. There were information from a random company very quickly and
more changes in betweenness centrality than in degree with less distortion (Borgatti et al., 2018). Once more, the
centrality. It means that there are more new companies in network configuration from one year to another is similar.
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 10
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 11
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
Only by seeing the graphs presented in Figures 2, 4, Some companies that were privileged in 2019 in terms of
6, and 7, is it possible to understand that the network in their power of influence have lost those positions to other
both years has a lower cohesion and a significant level of companies that have become more central in 2020.
fragmentation, with a high number of components, some
of which are with a small number of companies. Data All the steps taken throughout this tutorial, from
presented in Table 1 complement this qualitative evaluation, data collection, treatment, and possibilities of use are
informing each year’s density, number of components, summarized in Table 6. It is noteworthy that these are only
and the fragmentation and connectedness indexes. These a few examples of the potential for SNA in the investigation
measurements, along with the centralization, indicate few of the interlocks. Other properties and metrics can be
changes in network level analysis when comparing 2019 evaluated, compared, and/or used to analyze the effects of
with 2020. On the other hand, measurements at the level board diversity, Fiscal Council members’ networks, or other
of the nodes, especially the betweenness centrality, denote aspects from emerging economies, as suggested by Smith
significant changes in the positions of the companies. and Sarabi (2021b).
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 12
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
ADVANTAGES AND LIMITATIONS OF THE field of research. Furthermore, these measurements can be
METHOD used within the econometric models as variables in many
studies to identify board interlock precedents or outcomes.
Our tutorial has focused on networks formed between
The procedure provided in this tutorial regarding
companies that share their directors.
data collection, treatment, and social network analysis offers
several advantages over different and segregated procedures. Our main contribution to the proposition of this
First, data collection steps help researchers effectively use tutorial lies in providing a guideline so that future research can
GetFREData (Perlin et al., 2018), selecting the necessary be performed easily. Our primary goal has been to improve
data and the period covered to conduct the research. data reliability and orient researchers in the procedures to be
Second, data treatment procedures bypass the reduced conducted as well as automate most of the tasks that would
reliability problems created by data informed incorrectly be done by hand, subsequently reducing the time spent
by companies in the FRE. Third, all the steps required to treating and handling the dataset. Doing so, we believe this
conduct social network analysis regarding board interlocks tutorial can contribute to the enhancing of social network
are provided in this tutorial, making the process easier for all studies considering the Brazilian environment, increasing the
researchers to conduct and understand the networks formed understanding of board interlocks in an emergent country.
between companies that share at least one director. Finally, Santos, Silveira, and Barros (2007) have shown that interlocks
the empirical illustration of all procedures has demonstrated have been frequent in Brazilian publicly companies and
its advantages by providing an integrated and automatic that “larger boards, more dispersed ownership structures,
procedure for conducting the research. Before this, many and larger firm size are factors associated with higher level
steps were done separately and conducted mainly by hand. of board interlocking” (Santos, Silveira, & Barros, 2007, p.
Despite their advantages, for research into 126). However, quite a few studies have encompassed the
Brazilian public companies, these procedures have certain understanding of board interlocks’ inputs and outcomes in
limitations that provide avenues for further methodological the Brazilian context. Besides that, this tutorial contributes
development. First, we provided only a few adjustments in to the literature by providing a practical methodology to
name and CPF. Additional analyses could be done manually measure corporate governance indicators that can be related
to check directors’ names and CPFs, which could increase to the currently relevant ESG agenda. Our tutorial may
the number of observations. Moreover, as companies inform help researchers analyze, for example, the impact of board
the data without any filling pattern, it becomes hard to map connections on ESG outcomes, such as board diversity and
all the necessary procedures in order to guarantee complete CEO compensation.
reliability. Second, in our social network analysis steps, we Some avenues for future research using the procedures
have shown only some social network measures. For future described in this tutorial can be followed. Lamb and Roundy
research, there exists the opportunity to conduct a more (2016) suggest some research opportunities encompassing
in-depth analysis, draft out the steps and procedures to board interlocks, such as an evaluation of interlocks between
map out different networks, and include different network firms from different countries, an understanding of board
measurements, such as transitivity (Borgatti, 2009). interlock formation pattern (as firm age, for example), and
the relationships between interlocks and entrepreneurship.
CONCLUDING REMARKS Likewise, Sarabi and Smith (2021) suggest that although
there was an increase in the study of interlocked directors
The purpose of this tutorial article is to help researchers and its effects in recent years, the gender of the interlocked
conduct their research regarding social network analysis, directors is often neglected by researchers. In this sense, there
in particular board interlocks. To make this possible, we is an opportunity to deepen the knowledge of interlocks
have provided a step-by-step procedure to conduct data between women directors and how it impacts firms’ strategy
collection and data treatment. In addition, we showed the and decisions.
steps for building the adjacency matrix, generating and Additionally, this model is focused primarily on how
analyzing some network measurements and sociograms to collect and handle Brazilian publicly companies’ dataset
using R Studio. to be able to conduct social network analysis on board
To illustrate the process, we used a dataset with the interlocks. Although it can be adapted to be applied in
board of directors’ composition of Brazilian public companies different fields of studies, we acknowledge that all models
covering the years 2019 and 2020. We also analyzed network have restrictions and researchers can face necessity of
measurements to show how they can be interpreted in this adaptations to apply these instructions in other contexts.
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 13
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
REFERENCES
An, W., & Liu, Y.-H. (2016). Keyplayer: An R package for locating Davis, G. F., & Cobb, J. A. (2010). Resource dependence theory:
key players in social networks. The R Journal, 8(1), 257- Past and future. In F. Dobbin & C. B. Schoonhoven
268. https://doi.org/10.32614/RJ-2016-018 (Orgs.), Stanford’s organization theory renaissance, 1970-
2000 (pp. 21-42).Bingley, UK: Emerald Group.
Bebchuk, L. A., & Fried, J. M. (2004) Pay without performance: The
unfulfilled promise of executive compensation. Cambridge, Drago, C., Millo, F., Ricciuti, R., & Santella, P. (2015).
MA: Harvard University Press. Corporate governance reforms, interlocking
directorship and company performance in Italy.
Beuren, I. M., Pamplona, E., & Leite, M. (2020). Remuneração International Review of Law and Economics, 41, 38-49.
dos executivos e desempenho em empresas https://doi.org/10.1016/j.irle.2014.09.003
brasileiras familiares e não familiares. Revista de
Administração Contemporânea, 24(6), 514-531. Everett, M. G., & Borgatti, S. P. (2013). The dual-projection
https://doi.org/10.1590/1982-7849rac2020190191 approach for two-mode networks. Social Networks, 35(2),
Borgatti, S. P. (2009). Two-mode concepts in social network 204-210. https://doi.org/10.1016/j.socnet.2012.05.004
analysis. In R. Meyers (Ed.), Encyclopedia of complexity and Fich, E., & White, L. J. (2003). CEO compensation and turnover:
systems science (pp. 8279-8291). New York, NY: Springer. The effects of mutually interlocked boards. Wake Forest
https://doi.org/10.1007/978-0-387-30440-3_491 Law Review, 38, 935-959. Retrieved from https://
Borgatti, S. P., Mehra, A., Brass, D. J., & Labianca, G. (2009). heinonline.org/HOL/LandingPage?handle=hein.journals/
Network analysis in the social sciences. Science, 323(5916), wflr38&div=34&id=&page=
892-895. https://doi.org/10.1126/science.1165821 Fich, E., & White, L. J. (2005). Why do CEOs reciprocally sit on
Borgatti, S. P., Everett, M. G., & Johnson, J. C. (2018). Analyzing each other’s boards? Journal of Corporate Finance, 11(1-2),
social networks. Thousand Oaks, CA: SAGE. 175-195. https://doi.org/10.1016/j.jcorpfin.2003.06.002
Butts, C. T. (2008). Social network analysis with SNA. Freeman, L. C., Roeder, D., & Mulholland, R. R. (1979).
Journal of Statistical Software, 24(6), 1-51. Centrality in social networks: II. Experimental
https://doi.org/10.18637/jss.v024.i06 results. Social Networks, 2(2), 119-141.
https://doi.org/10.1016/0378-8733(79)90002-9
Certo, S. T. (2003). Influencing initial public offering investors
with prestige: Signaling with board structures. Hamdan, A. (2018), Board interlocking and firm performance: The
Academy of Management Review, 28(3), 432-446. role of foreign ownership in Saudi Arabia. International
https://doi.org/10.5465/amr.2003.10196754 Journal of Managerial Finance, 14(3), 266-281.
https://doi.org/10.1108/IJMF-09-2017-0192
Costa, C. M., Galdi, F. C., Motoki, F. Y., & Sanchez, J. M.
(2016). Non‐compliance in executive compensation Harjoto, M. A., & Wang, Y. (2020). Board of directors network
disclosure: The Brazilian experience. Journal of centrality and environmental, social and governance
Business Finance & Accounting, 43(3-4), 329-369. (ESG) performance. Corporate Governance, 20(6), 965-
https://doi.org/10.1111/jbfa.12177 985. https://doi.org/10.1108/CG-10-2019-0306
Crabtree, J. (2011, May 10). One woman, multiple boards: Rise of Haunschild, P. R., & Beckman, C. M. (1998), When do interlocks
the ‘super-connected’ director. Financial Times. Retrieved matter? Alternate sources of information and interlock
from https://www.ft.com/content/620ad5a0-7a20-11e0- influence. Administrative Science Quarterly, 43(4), 815-
bc74-00144feabdc0 844. https://doi.org/10.2307/2393617
Croci, E., & Grassi, R. (2014). The economic effect of Helmers, C., Patnam, M., & Rau, P. R. (2017). Do board interlocks
interlocking directorates in Italy: New evidence increase innovation? Evidence from a corporate governance
using centrality measures. Computational and reform in India. Journal of Banking & Finance, 80, 51-70.
Mathematical Organization Theory, 20(1), 89–112. https://doi.org/10.1016/j.jbankfin.2017.04.001
https://doi.org/10.1007/s10588-013-9154-1
Higgins, M. C., & Gulati, R. (2003). Getting off to a good start:
Csardi, G., & Nepusz, T. (2006). The igraph software package for The effects of upper echelon affiliations on underwriter
complex network research. InterJournal, Complex Systems, prestige. Organization Science, 14(3), 244-263. Retrieved
1695(5), 1-9. from https://www.jstor.org/stable/4135135
Comissão de Valores Mobiliários. (2009). Instrução CVM 480 - Hillman, A. J., Withers, M. C., & Collins, B. J. (2009).
Dispõe sobre o registro de emissores de valores mobiliários Resource dependence theory: A review.
admitidos à negociação em mercados regulamentados de Journal of Management, 35(6), 1404-1427.
valores mobiliários. Rio de Janeiro, RJ, Brasil. Retrieved https://doi.org/10.1177%2F0149206309343469
from https://conteudo.cvm.gov.br/legislacao/instrucoes/
inst480.html#:~:text=Disp%C3%B5e%20sobre%20 Kang, E. (2008). Director interlocks and spillover effects of
o%20registro%20de,134%2F90%3B%20o%20art reputational penalties from financial reporting fraud.
Academy of Management Journal, 51(3), 537-555.
Dahlin, P., & Patel, P. C. (2022). Joined by remoteness: An https://doi.org/10.5465/amj.2008.32626007
exploratory comparison of regional board networks
in Sweden. Global Networks, 22(2), 325-343. Kilduff, M., & Tsai, W. (2003). Social networks and organizations.
https://doi.org/10.1111/glob.12347 London: SAGE Publications.
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 14
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
Lamb, N. H., & Roundy, P. (2016). The “ties that bind” Scott, J. (1992). Corporate interlocks and social network analysis.
board interlocks research: A systematic review. Hong Kong: Social Sciences Research Centre of
Management Research Review, 39(11), 1516-1542. the University of Hong Kong. Retrieved from
https://doi.org/10.1108/MRR-02-2015-0027 http://hdl.handle.net/10722/42551
Larcker, D. F., Richardson, S. A., Seary, A., & Tuna, A. (2005). Back Silva, L. K. S., Silva, A. R. H., Vasconcelos, A. C., & Crisóstomo,
door links between directors and executive compensation. V. L. (2019, July). Influência do board interlocking e do
SSRN. https://doi.org/10.2139/ssrn.671063
controle acionário na qualidade da governança corporativa
Lazzarini, S. G. (2008). Empresas em rede. São Paulo: Cengage no mercado de ações Brasileiro. Proceedings of the USP
Learning. International Conference in Accounting, São Paulo, SP,
Brazil, 29. Retrieved from https://congressousp.fipecafi.
Lei n. 6.404, de 15 de dezembro de 1976 (1976). Dispõe org/anais/19UspInternational/ArtigosDownload/1721.
sobre as sociedades por ações. Diário Oficial da
União. Brasília, DF: Casa Civil. Retrieved from pdf
http://www.planalto.gov.br/ccivil_03/leis/l6404consol.htm Smith, M., & Sarabi, Y. (2021a). UK trading patterns within
Locatelli, L. G., Ramos, F. M., & Costa, C. M. (2021). and between regions in the automotive sector — A
Conexões sociais e rotatividade involuntária do network analysis. The World Economy, 44(2), 510-529.
CEO: Evidências do mercado brasileiro. Revista https://doi.org/10.1111/twec.13006
Contemporânea de Contabilidade, 18(48), 124-137.
https://doi.org/10.5007/2175-8069.2021.e76116 Smith, M., & Sarabi, Y. (2021b). “What do interlocks do” revisited – A
bibliometric analysis. Management Research Review, 44(4),
Mastella, M., Vancin, D., Perlin, M., & Kirch, G. (2021). Board 642-659. https://doi.org/10.1108/MRR-05-2020-0258
gender diversity: Performance and risk of Brazilian
firms. Gender in Management, 36(4), 498-518. Sprenger, K. B., Kronbauer, C. A., & Costa, C. M. (2017).
https://doi.org/10.1108/GM-06-2019-0088 Características do CEO e o gerenciamento de resultados em
empresas listadas na BM&FBOVESPA. Revista Universo
Mazzola, E., Perrone, G., & Kamuriwo, D. S. (2016). The Contábil, 13(3), 120-142. Retrieved from https://proxy.
interaction between inter-firm and interlocking
directorate networks on firm’s new product development furb.br/ojs/index.php/universocontabil/article/view/6425
outcomes. Journal of Business Research, 69(2), 672-682. Tham, Y. H., Sultana, N., Singh, H., & Taplin, R. (2019). Busy
https://doi.org/10.1016/j.jbusres.2015.08.033 boards and earnings management – an Australian
Meghanathan, N. (2017). Graph theoretic approaches for analyzing perspective. Asian Review of Accounting, 27(3), 464-486.
large-scale social networks. Hershey, PA: IGI Global. https://doi.org/10.1108/ARA-08-2018-0149
https://doi.org/10.4018/978-1-5225-2814-2
Useem, M. (1986), The inner circle: Large corporations and the rise
Perlin, M., Kirch, G., & Vancin, D. (2018). Accessing financial of business political activity in the US and UK. Oxford:
reports and corporate events with GetDFPData. SSRN. Oxford University Press.
https://doi.org/10.2139/ssrn.3128252
Vesco, D. G. D., & Beuren, I. M. (2016). Do the
Pfeffer, J. (1983), Organizational demography. In L. L. Cummings board of directors’ composition and the board
& B. M. Staw (Orgs.). Research in Organizational Behavior interlocking influence on performance? BAR -
(pp. 299-357). Greenwich, CT: JAI Press. Brazilian Administration Review, 13(2), e160007.
Phan, P. H., Lee, S. H., & Lau, S. C. (2003). The performance https://doi.org/10.1590/1807-7692bar2016160007
impact of interlocking directorates: The case of Singapore.
Journal of Managerial Issues, 15(3) 338-352. Retrieved Yeo, H.-J., Pochet, C., & Alcouffe, A. (2003). CEO
from https://www.jstor.org/stable/40604436 reciprocal interlocks in French corporations. Journal
of Management and Governance, 7(1), 87-108.
Rocha, L. E., Ryckebusch, J., Schoors, K., & Smith, M. https://doi.org/10.1023/A:1022442602193
(2021). The scaling of social interactions across
animal species. Scientific Reports, 11(1), 12584. Wasserman, S., & Faust, K. (1994). Social network analysis: Methods
https://doi.org/10.1038/s41598-021-92025-1 and applications. Cambridge: Cambridge University Press.
https://doi.org/10.1017/CBO9780511815478
Rodrigues, S. D. S., & Galdi, F. C. (2017). Investor
relations and information asymmetry. Revista Wong, L. H. H., Gygax, A. F., & Wang, P. (2015). Board
Contabilidade & Finanças, 28(74), 297-312. interlocking network and the design of executive
https://doi.org/10.1590/1808-057x201703630 compensation packages. Social Networks, 41, 85-100.
Santos, R. L., da Silveira, A. D. M., & Barros, L. A. B. D. C. (2007). https://doi.org/10.1016/j.socnet.2014.12.002
Board interlocking in Brazil: Director participation
in multiple companies and its effect on the value of Wu, Y., Zhang, Y., & Chen, Y. (2021). The influence of board
firms. Revista Brasileira de Finanças, 5(2), 125-163. interlocking network centrality on foundation performance:
https://doi.org/10.12660/rbfin.v5n2.2007.1173 Evidence from China. Journal of Chinese Governance, 6(1),
43-57. https://doi.org/10.1080/23812346.2020.1841483
Sarabi, Y., & Smith, M. (2021). Busy female directors: An
exploratory analysis of the impact of quotas and interest Zhu, B., Watts, S., & Chen, H. (2010). Visualizing social network
groups. Gender in Management, 36(3), 368-385. concepts. Decision Support Systems, 49(2), 151-161.
https://doi.org/10.1108/gm-07-2019-0129 https://doi.org/10.1016/j.dss.2010.02.001
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 15
Board of directors’ interlocks: A social network analysis tutorial C. P. Salgado, V. S. Adami, J. R. de S. Verschoore Filho, C. M. Costa
RAC is a member of, and subscribes to the principles of the Committee on Publication Ethics (COPE) for scholarly publication
Revista de Administração Contemporânea, 2022, Early access | doi.org/10.1590/1982-7849rac2022210286.en| e-ISSN 1982-7849 | rac.anpad.org.br 16